StackAHTPs: An explainable antihypertensive peptides identifier based on heterogeneous features and stacked learning...

Area of Science:

• Biotechnology
• Bioinformatics
• Computational Biology

Background:

• Hypertension (high blood pressure) affects millions globally, increasing cardiovascular disease risk.
• Naturally derived bioactive peptides show promise for reducing blood pressure, offering alternatives to pharmaceuticals.
• Traditional methods for identifying antihypertensive peptides (AHTPs) are costly and time-consuming.

Purpose of the Study:

• To develop a novel, accurate, and efficient in-silico method for predicting antihypertensive peptides (AHTPs) using only sequence data.
• To accelerate the drug discovery process for novel antihypertensive agents.

Main Methods:

• Developed StackAHTP, a machine-learning predictor utilizing Pseudo-Amino Acid Composition and Dipeptide Composition for feature extraction.
• Employed SHapley Additive explanations (SHAP) for feature ranking and ensemble classifiers (Bagging, Boosting, Stacking) for enhanced prediction.
• Validated the model using 10-fold cross-validation and an independent test set.

Main Results:

• The StackAHTP model achieved high prediction accuracy (92.25%) and F1-score (89.67%) on an independent test set.
• Outperformed existing machine-learning classifiers including AdaBoost, XGBoost, and LightGBM.
• Demonstrated the effectiveness of combining sequence-based features and ensemble methods for AHTP prediction.

Conclusions:

• The developed StackAHTP predictor offers a cost-effective and time-efficient in-silico approach for identifying potential antihypertensive peptides.
• This research significantly contributes to the large-scale characterization and accelerated discovery of AHTPs.
• The study provides a valuable tool for researchers in drug discovery and hypertension management.